Bridge over troubled water: the apparent discrepancy between simulated and experimental non-ambient water structure

Publisher Website
Google Scholar
Abstract
Recent neutron diffraction experiments, which exploit hydrogen isotope substitution techniques to extract the HH, OH and OO site - site radial distribution functions for water, indicate that as the temperature of water is raised above the critical point, the hydrogen-bonding network, as measured by the height of the first peak in the OH distribution function, collapses. Several computer simulations, however, dispute the accuracy of the experimentally determined distribution functions: they show that the measured data cannot be obtained from any physical arrangement of water molecules. By applying additional constraints on the small-radius behaviour of the radial distribution functions, and by repeating some of the experiments under different experimental conditions, the accuracy of the extracted radial distribution functions has now been improved. The new distribution functions, while not qualitatively different from what has already been published, satisfy the fundamental objection to the previous results, namely they can be simulated with physical assemblies of water molecules. These simulated distributions of molecules indicate a weak degree of hydrogen bonding in water at 673 K which is greatly reduced compared to that in the hydrogen-bond network of ambient water. The analysis is supported by a new computer simulation of water structure using an `empirical' water interaction potential, which contains none of the long-range features of traditional charge models, such as SPC/E. This short-range potential is able to reproduce most features of the experimental data to good accuracy, even under supercritical conditions.